From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail.free-electrons.com ([62.4.15.54]) by bombadil.infradead.org with esmtp (Exim 4.89 #1 (Red Hat Linux)) id 1ebu2b-0004T9-K3 for linux-mtd@lists.infradead.org; Wed, 17 Jan 2018 20:10:36 +0000 Date: Wed, 17 Jan 2018 21:10:18 +0100 From: Boris Brezillon To: Ezequiel Garcia Cc: Richard Weinberger , Bradley Bolen , linux-mtd@lists.infradead.org, Artem Bityutskiy , Marek Vasut , Cyrille Pitchen , Brian Norris , David Woodhouse Subject: Re: [PATCH] UBI: block: Must use a mutex when using idr_alloc/idr_remove Message-ID: <20180117211018.4327c9d4@bbrezillon> In-Reply-To: References: <1514946369-3674-1-git-send-email-bradleybolen@gmail.com> <5aaad0f3-2d7d-2c1f-3ad1-6049a5a0c088@gmail.com> <2145261.UlTadXsT1L@blindfold> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wed, 17 Jan 2018 17:06:29 -0300 Ezequiel Garcia wrote: > On 17 January 2018 at 17:04, Richard Weinberger wrote: > > Am Mittwoch, 17. Januar 2018, 20:46:51 CET schrieb Bradley Bolen: > >> On 01/14/2018 05:20 PM, Ezequiel Garcia wrote: > >> > On 14 January 2018 at 11:39, Boris Brezillon > >> > > >> > wrote: > >> >> +Ezequiel > >> >> > >> >> Hi Bradley, > >> >> > >> >> On Tue, 2 Jan 2018 21:26:09 -0500 > >> >> > >> >> Bradley Bolen wrote: > >> >>> This fixes a race condition where running ubiblock on multiple volumes > >> >>> simultaneously produces the following splat. > >> >>> > >> >>> kernel BUG at kernel-source/fs/sysfs/group.c:113! > >> >>> Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM > >> >>> [] (internal_create_group) from [] > >> >>> (sysfs_create_group+0x20/0x24) > >> >>> [] (sysfs_create_group) from [] > >> >>> (blk_trace_init_sysfs+0x18/0x20) > >> >>> [] (blk_trace_init_sysfs) from [] > >> >>> (blk_register_queue+0x6c/0x154) > >> >>> [] (blk_register_queue) from [] > >> >>> (device_add_disk+0x2c8/0x450) > >> >>> [] (device_add_disk) from [] > >> >>> (ubiblock_create+0x254/0x2e4) > >> >>> [] (ubiblock_create) from [] > >> >>> (vol_cdev_ioctl+0x3e0/0x564) > >> >>> [] (vol_cdev_ioctl) from [] (vfs_ioctl+0x30/0x44) > >> >>> [] (vfs_ioctl) from [] (do_vfs_ioctl+0x694/0x798) > >> >>> [] (do_vfs_ioctl) from [] (SyS_ioctl+0x44/0x6c) > >> >>> [] (SyS_ioctl) from [] (ret_fast_syscall+0x0/0x34) > >> > > >> > I'm a little confused, how does this stack trace relates to idr? > >> > >> I apologize. The commit message was a little terse. Here is the full > >> output of the issue I ran into. > >> > >> ------------[ cut here ]------------ > >> block ubiblock0_3: created from ubi0:3(Partition3) > >> block ubiblock0_2: created from ubi0:2(Partition2) > >> block ubiblock0_4: created from ubi0:4(Partition4) > >> WARNING: CPU: 1 PID: 179 at kernel-source/fs/sysfs/dir.c:31 > >> sysfs_warn_dup+0x68/0x88 sysfs: cannot create duplicate filename > >> '/devices/virtual/bdi/252:2' Modules linked in: > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> [] (unwind_backtrace) from [] (show_stack+0x20/0x24) > >> [] (show_stack) from [] (dump_stack+0x78/0x94) > >> [] (dump_stack) from [] (__warn+0xf0/0x110) > >> [] (__warn) from [] (warn_slowpath_fmt+0x54/0x74) > >> [] (warn_slowpath_fmt) from [] > >> (sysfs_warn_dup+0x68/0x88) [] (sysfs_warn_dup) from [] > >> (sysfs_create_dir_ns+0x84/0x94) [] (sysfs_create_dir_ns) from > >> [] (kobject_add_internal+0x130/0x2f8) [] > >> (kobject_add_internal) from [] (kobject_add+0x9c/0xb8) > >> [] (kobject_add) from [] (device_add+0xfc/0x56c) > >> [] (device_add) from [] > >> (device_create_groups_vargs+0x90/0xd0) [] > >> (device_create_groups_vargs) from [] > >> (device_create_vargs+0x34/0x3c) [] (device_create_vargs) from > >> [] (bdi_register+0x70/0x1cc) [] (bdi_register) from > >> [] (bdi_register_owner+0x3c/0x7c) [] > >> (bdi_register_owner) from [] (device_add_disk+0x114/0x44c) > >> [] (device_add_disk) from [] > >> (ubiblock_create+0x284/0x2e0) [] (ubiblock_create) from > >> [] (vol_cdev_ioctl+0x450/0x554) [] (vol_cdev_ioctl) > >> from [] (vfs_ioctl+0x30/0x44) [] (vfs_ioctl) from > >> [] (do_vfs_ioctl+0xa0/0x790) [] (do_vfs_ioctl) from > >> [] (SyS_ioctl+0x44/0x68) [] (SyS_ioctl) from > >> [] (ret_fast_syscall+0x0/0x34) ---[ end trace e541910253daa337 > >> ]--- > >> ------------[ cut here ]------------ > >> WARNING: CPU: 1 PID: 179 at kernel-source/lib/kobject.c:240 > >> kobject_add_internal+0x1ec/0x2f8 kobject_add_internal failed for 252:2 with > >> -EEXIST, don't try to register things with the same name in the same direc > >> tory. > >> Modules linked in: > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P W O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> [] (unwind_backtrace) from [] (show_stack+0x20/0x24) > >> [] (show_stack) from [] (dump_stack+0x78/0x94) > >> [] (dump_stack) from [] (__warn+0xf0/0x110) > >> [] (__warn) from [] (warn_slowpath_fmt+0x54/0x74) > >> [] (warn_slowpath_fmt) from [] > >> (kobject_add_internal+0x1ec/0x2f8) [] (kobject_add_internal) from > >> [] (kobject_add+0x9c/0xb8) [] (kobject_add) from > >> [] (device_add+0xfc/0x56c) [] (device_add) from > >> [] (device_create_groups_vargs+0x90/0xd0) [] > >> (device_create_groups_vargs) from [] > >> (device_create_vargs+0x34/0x3c) [] (device_create_vargs) from > >> [] (bdi_register+0x70/0x1cc) [] (bdi_register) from > >> [] (bdi_register_owner+0x3c/0x7c) [] > >> (bdi_register_owner) from [] (device_add_disk+0x114/0x44c) > >> [] (device_add_disk) from [] > >> (ubiblock_create+0x284/0x2e0) [] (ubiblock_create) from > >> [] (vol_cdev_ioctl+0x450/0x554) [] (vol_cdev_ioctl) > >> from [] (vfs_ioctl+0x30/0x44) [] (vfs_ioctl) from > >> [] (do_vfs_ioctl+0xa0/0x790) [] (do_vfs_ioctl) from > >> [] (SyS_ioctl+0x44/0x68) [] (SyS_ioctl) from > >> [] (ret_fast_syscall+0x0/0x34) ---[ end trace e541910253daa338 > >> ]--- > >> ------------[ cut here ]------------ > >> WARNING: CPU: 1 PID: 179 at kernel-source/fs/sysfs/dir.c:31 > >> sysfs_warn_dup+0x68/0x88 sysfs: cannot create duplicate filename > >> '/dev/block/252:2' > >> Modules linked in: > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P W O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> [] (unwind_backtrace) from [] (show_stack+0x20/0x24) > >> [] (show_stack) from [] (dump_stack+0x78/0x94) > >> [] (dump_stack) from [] (__warn+0xf0/0x110) > >> [] (__warn) from [] (warn_slowpath_fmt+0x54/0x74) > >> [] (warn_slowpath_fmt) from [] > >> (sysfs_warn_dup+0x68/0x88) [] (sysfs_warn_dup) from [] > >> (sysfs_do_create_link_sd+0xa0/0xbc) [] (sysfs_do_create_link_sd) > >> from [] (sysfs_create_link+0x34/0x44) [] > >> (sysfs_create_link) from [] (device_add+0x42c/0x56c) [] > >> (device_add) from [] (device_add_disk+0x160/0x44c) [] > >> (device_add_disk) from [] (ubiblock_create+0x284/0x2e0) > >> [] (ubiblock_create) from [] > >> (vol_cdev_ioctl+0x450/0x554) [] (vol_cdev_ioctl) from > >> [] (vfs_ioctl+0x30/0x44) [] (vfs_ioctl) from > >> [] (do_vfs_ioctl+0xa0/0x790) [] (do_vfs_ioctl) from > >> [] (SyS_ioctl+0x44/0x68) [] (SyS_ioctl) from > >> [] (ret_fast_syscall+0x0/0x34) ---[ end trace e541910253daa339 > >> ]--- > >> ubiblock (179): undefined instruction: pc=c01e26cc > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P W O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> task: c185d780 task.stack: d053a000 > >> PC is at internal_create_group+0x3c/0x2a0 > >> LR is at sysfs_create_group+0x20/0x24 > >> pc : [] lr : [] psr: 60000013 > >> sp : d053bd38 ip : d053bd70 fp : d053bd6c > >> r10: c095fddc r9 : 00000000 r8 : c1b59c00 > >> r7 : c1b59870 r6 : c090f488 r5 : c09276f8 r4 : 00000000 > >> r3 : 00000000 r2 : c09276f8 r1 : 00000000 r0 : c1b59870 > >> Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user > >> Control: 30c5383d Table: 104b6840 DAC: ffab73de > >> Code: e89da8f0 e1a0c00d e92ddff0 e24cb004 e24dd00c e52de004 e8bd4000 > >> e2507000 e1a09001 e1a05002 0a000004 e3510000 e59 74018 1a000002 e3540000 > >> 1a000002 (e7f001f2) e3540000 0a00000f e595300c e5951000 e3530000 1a00000d > >> e5953010 e3530000 1 a00000a e59f2220 e3510000 e5973000 e59f0218 01a01002 > >> e59f2214 > >> ------------[ cut here ]------------ > >> kernel BUG at kernel-source/fs/sysfs/group.c:113! > >> Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM > >> -----------[ user debug ]----------- > >> Current process: ubiblock pid=179 > >> Running on CPU1 (2 online) > >> > >> current=c185d780 preempt_count=0x0 > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P W O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> task: c185d780 task.stack: d053a000 > >> PC is at 0x4b767d8c > >> LR is at 0x10b78 > >> pc : [<4b767d8c>] lr : [<00010b78>] psr: 60000010 > >> sp : be73fc34 ip : 4b767d80 fp : 00000000 > >> r10: 00000003 r9 : be73fdb4 r8 : 00000000 > >> r7 : 00000036 r6 : 00000003 r5 : 000250a8 r4 : ffffffff > >> r3 : 00000001 r2 : 22eea600 r1 : 40804f07 r0 : 00000003 > >> Flags: nZCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user > >> Control: 30c5383d Table: 104b6840 DAC: ffab73de > >> > >> Memory Maps: > >> 00010- > >> 00015 > >> ubiblock > >> > >> 00024- > >> 00026 > >> ubiblock > >> > >> 00026- > >> 00047 > >> > >> 4b660- > >> 4b67f > >> ld-2 > >> > >> 4b68e- > >> 4b690 > >> ld-2 > >> > >> 4b6a0- > >> 4b7da > >> libc-2 > >> > >> 4b7da- > >> 4b7dc > >> > >> b6bad- > >> b6baf > >> > >> be71f- > >> be740 > >> > >> beb84- > >> ---------[ end user debug ]--------- > >> Modules linked in: > >> CPU: 1 PID: 179 Comm: ubiblock Tainted: P W O > >> 4.11.12-yocto-standard-414e5550071e0ca8639b0d8fb84fed31 #1 Hardware name: > >> (Device Tree) > >> task: c185d780 task.stack: d053a000 > >> PC is at internal_create_group+0x3c/0x2a0 > >> LR is at sysfs_create_group+0x20/0x24 > >> pc : [] lr : [] psr: 60000013 > >> sp : d053bd38 ip : d053bd70 fp : d053bd6c > >> r10: c095fddc r9 : 00000000 r8 : c1b59c00 > >> r7 : c1b59870 r6 : c090f488 r5 : c09276f8 r4 : 00000000 > >> r3 : 00000000 r2 : c09276f8 r1 : 00000000 r0 : c1b59870 > >> Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user > >> Control: 30c5383d Table: 104b6840 DAC: ffab73de > >> Process ubiblock (pid: 179, stack limit = 0xd053a210) > >> Stack: (0xd053bd38 to 0xd053c000) > >> bd20: d053bd5c > >> d053bd48 bd40: c00731c8 d0bd9e00 c1b59800 c090f488 c1b59868 c1b59c00 > >> c1b5980c c095fddc bd60: d053bd7c d053bd70 c01e2950 c01e269c d053bd8c > >> d053bd80 c00e3d38 c01e293c bd80: d053bdbc d053bd90 c02bdfbc c00e3d2c > >> 00040b0c c1b59800 c1b59868 c090f488 bda0: c1b59870 c1b59c00 c1b5980c > >> c095fddc d053be14 d053bdc0 c02cec84 c02bdef0 bdc0: c02ce6cc c1b59800 > >> d053bdfc d053bdd8 0fc00002 c0053478 00000000 1f518000 bde0: d053be1c > >> 00040b0c 00000000 c1b11e40 00000000 c1b59800 c095fda4 d053be60 be00: > >> c1b11e8c c095fddc d053be54 d053be18 c0436ec8 c02ceafc 00000000 c1b5980c > >> be20: c01e06f0 c1b5980c 00000000 22eea600 d04b6100 d0b7cc00 d1c20000 > >> c090f488 be40: d053a000 c090f488 d053bedc d053be58 c0427bb8 c0436c50 > >> 00000001 00000000 be60: 00000000 0000000f 00000017 00000003 002c9000 > >> 00000000 00000001 00000003 be80: 00000000 00000000 00000001 0001f000 > >> 00000006 d0b7ce5c 0f700010 c005368c bea0: 9a877d41 00000007 c1aab015 > >> c0755170 00000261 00040b0c d053beec 22eea600 bec0: d0b96228 c1896300 > >> 00000003 22eea600 d053beec d053bee0 c0189110 c0427774 bee0: d053bf7c > >> d053bef0 c01892e0 c01890ec d053bf34 c0185224 c01954e8 c05e9a14 bf00: > >> c0185278 c1aab000 c090f488 c1896300 d0b96228 c1aab000 c1896308 00000020 > >> bf20: d053bf44 d053bf30 c0185224 c0163200 00000003 c090f488 d053bf94 > >> d053bf48 bf40: c0175504 c01851c4 00000000 00040b0c 00000002 00000003 > >> c1896300 c1896300 bf60: 40804f07 22eea600 d053a000 00000000 d053bfa4 > >> d053bf80 c0189a14 c018924c bf80: ffffffff 000250a8 00000003 00000036 > >> c00107e4 d053a000 00000000 d053bfa8 bfa0: c0010640 c01899dc ffffffff > >> 000250a8 00000003 40804f07 22eea600 00000001 bfc0: ffffffff 000250a8 > >> 00000003 00000036 00000000 be73fdb4 00000003 00000000 bfe0: 4b767d80 > >> be73fc34 00010b78 4b767d8c 60000010 00000003 e675ff3a 2a3139dd [] > >> (internal_create_group) from [] (sysfs_create_group+0x20/0x24) > >> [] (sysfs_create_group) from [] > >> (blk_trace_init_sysfs+0x18/0x20) [] (blk_trace_init_sysfs) from > >> [] (blk_register_queue+0xd8/0x154) [] > >> (blk_register_queue) from [] (device_add_disk+0x194/0x44c) > >> [] (device_add_disk) from [] > >> (ubiblock_create+0x284/0x2e0) [] (ubiblock_create) from > >> [] (vol_cdev_ioctl+0x450/0x554) [] (vol_cdev_ioctl) > >> from [] (vfs_ioctl+0x30/0x44) [] (vfs_ioctl) from > >> [] (do_vfs_ioctl+0xa0/0x790) [] (do_vfs_ioctl) from > >> [] (SyS_ioctl+0x44/0x68) [] (SyS_ioctl) from > >> [] (ret_fast_syscall+0x0/0x34) Code: e89da8f0 e1a0c00d e92ddff0 > >> e24cb004 e24dd00c e52de004 e8bd4000 e2507000 e1a09001 e1a05002 0a000004 > >> e3510000 e59 74018 1a000002 e3540000 1a000002 (e7f001f2) e3540000 0a00000f > >> e595300c e5951000 e3530000 1a00000d e5953010 e3530000 1 a00000a e59f2220 > >> e3510000 e5973000 e59f0218 01a01002 e59f2214 > >> ---[ end trace e541910253daa33a ]--- > >> > >> I was able to reproduce the problem with the following script with the > >> underlying volumes already created: > >> > >> while true; do > >> for part in 2 3 4 15 24 90 91; do > >> ubiblock -c /dev/ubi0_$part & > >> done > >> sleep 2 > >> > >> for part in 2 3 4 15 24 90 91; do > >> ubiblock -r /dev/ubi0_$part & > >> done > >> sleep 2 > >> done > >> > >> There is a race with idr_alloc where gd->first_minor could be set to the > >> same value for two different calls to ubiblock_create. Each instance > >> calls device_add_disk with the same first_minor. device_add_disk calls > >> bdi_register_owner. The bdi_register_owner call path produces the > >> initial WARNINGS about registering the same device. However, > >> device_add_disk does not error out when bdi_register_owner returns an > >> error. The control continues until it reaches blk_register_queue which > >> eventually BUGs. > >> > >> >> Missing Fixes and Cc-stable tags. > >> >> > >> >>> Signed-off-by: Bradley Bolen > >> >> > >> >> Reviewed-by: Boris Brezillon > >> >> > >> >> BTW, just had a quick look at the code, and I think the locking can be > >> >> simplified a bit by keeping the devices_lock for the whole > >> >> ubiblock_create() operation instead of acquiring/releasing it several > >> >> times in the function (see [1]). It makes sense to do fine grained > >> >> locking when operations protected by the lock are time sensitive, but I > >> >> don't think this is the case here. > >> > > >> > Agreed. Something like this perhaps http://ix.io/E8G ? > >> > Notice that ubiblock_remove_all seemed to require locking as well. > >> > >> I tried out Ezequiel's patch and it fixes this issue as well. Note, > >> that ret needs to be defined in ubiblock_remove. > > > > Can you please have a proper patch such that I can take it? :-) > > > > Bradley, > > Feel free to submit a patch based on the (untested) diff I sent. > While there, you can improve the commit log with the details > you sent. Can we fix the locking in ubiblock_remove_all() as well?